APA Dictionary of Psychology

n. the ability of a statistical procedure to produce valid estimated values for a population characteristic (parameter) despite violations of the assumptions on which the procedure is based.